National MDS Study: Targeted Sequencing of Seven Genes Can Help Reduce Misclassification

According to research presented at the 62nd ASH Annual Meeting & Exposition, combining pathology reviews with targeted sequencing of seven particular genes could improve diagnosis of myelodysplastic syndromes (MDS). The study used targeted gene sequencing data to refine classification and increase agreement in diagnosis of MDS versus “other.”

The research was conducted as part of the National MDS Study, a project of the National Heart, Lung, and Blood Institute, part of the National Institutes of Health. It involves 92 community hospitals and 29 academic centers. Together, they are working to establish a cohort of 2,000 adults who have been recently diagnosed with MDS as well as 500 adults with idiopathic cytopenia of undetermined significance. The goals are to elucidate the genetic, epigenetic, and biological factors associated with initiation and progression of MDS.

The National MDS Study collects bone marrow and blood as well as information about clinical characteristics, such as:

• Symptoms
• Quality of life
• Treatments
• Severity of illness over time

Each patient undergoes local and centralized pathology review and is assigned a diagnosis:

• MDS
• Myelodysplastic/myeloproliferative neoplasms
• Acute myeloid leukemia with blasts <30%
• “Other”

Central pathologists examined patients’ peripheral blood and bone marrow biopsy specimens, clinical history, complete blood count, karyotyping, fluorescence in situ hybridization, and pathology reports from local pathologists. They used the updated 2016 World Health Organization classifications to diagnose MDS. Then, with bone marrow specimens, the researchers used targeted exon sequencing of 96 genes.

Analysis involved 648 individuals classified as MDS (n=212) or other (n=436). Patients diagnosed with MDS or other by both the central and the local pathologists were the “gold standard” (GS) (n=546). Seventy-five percent of the GS cases were used to train and validate the models, and the remaining 25% of GS cases were used to identify an optimal probability cutoff point that could be used to classifying subjects as having MDS.

That model was then used on other patients, including a group who had disagreement from local and central pathologists. The process revealed seven genes that were most helpful in differentiating MDS from “other”:

• TP53
• SF3B1
• U2AF1
• ASXL1
• TET2
• STAG2
• SRSF2

That model was then applied to the 50 patients who had disagreement from local and central pathologists, and the model accurately reassigned, with agreement, 37 of those subjects. Overall, the study found that three of 16 “other” cases had been misclassified as MDS by the local pathologist.

When the researchers assessed the model’s ability to predict MDS versus “other” for another group of 52 subjects, overall accuracy was 83%.